Julia Releases

A compact index of HPR-LP (Julia) releases—each entry shows the version tag and a one-line summary; click a version to view full notes, benchmarks, and downloads.

v0.1.3 — Robustness upgrade: refined parameter schedule achieves 1e-9 KKT/duality gaps and streamlines JuMP modeling

Release date: 2025-10-17

Highlights

  1. Enhanced parameter adjustment strategy for significantly improved stability, achieving relative KKT and duality gap accuracy up to 1e-9.

  2. Improved LP modeling pipeline with seamless JuMP integration for a smoother modeling experience.

Benchmark results

  • Platform: NVIDIA A100-SXM4-80GB

  • Dataset: Mittelmann’s LP benchmark (no presolve)

  • Performance: 47 / 49 instances solved (Tolerance: 1e-4, Time limit: 3600s)

  • Performance: 41 / 49 instances solved (Tolerance: 1e-9, Time limit: 3600s)

Downloads

v0.1.2 — SpMV & kernel rewrites

Release date: 2025-09-27

Highlights

  • SpMV rewrites: added preprocessing and buffer preallocation to avoid redundant work between iterations.

  • CUDA kernel refactors: reduced memory traffic and improved occupancy.

  • Under SGM10 (1e-8 accuracy): +11% on Mittelmann’s LP benchmark set and +7% on MIP2017 large-scale LP relaxations (vs. v0.1.1).

Downloads

v0.1.1 — Model reformulation + adaptive restart/penalty + fused kernels

Release date: 2025-09-09

Highlights

  • Model reformulation. Updated the problem formulation to the new form for better stability and consistency across instances.

  • Adaptive restarts & penalty auto-tuning. A redesigned penalty parameter update rule to improve convergence speed and robustness.

  • Kernel rewrites. Several CUDA kernels were refactored/fused to reduce memory traffic and improve occupancy.

  • Simplified parameters. Removed sigma and sigma_fixed from the parameters.

  • In terms of SGM10 (1e-8 accuracy), 14% faster for Mittelmann’s LP benchmark set and 95% faster for MIP2017 large-scale LP relaxations (compared to v0.1.0).

Downloads

v0.1.0 — A preliminary release: Julia GPU-accelerated HPR-LP solver

Release date: 2025-07-04

Highlights

  • First public release of HPR-LP.

  • GPU-accelerated LP solver in Julia implementing the Halpern–Peaceman–Rachford (HPR) method.

  • Model formulation:

\[\begin{split}\begin{aligned} \min_{x \in \mathbb{R}^n} \quad & (c, x) \\ \text{s.t.}\quad & A_1 x = b_1, \\ & A_2 x \ge b_2, \\ & l \le x \le u. \end{aligned}\end{split}\]

Downloads